High-density oligonucleotide arrays provide a means to monitor the expression levels of tens of thousands of genes simultaneously. The massive amount of data generated by this technology will be a key source of information that will help to clarify the roles, functions and regulation of the large number of genes identified in the Human Genome Project. The long-term objective of this project is to develop statistical methods and computer programs for effective analysis of gene expression data from high-density oligonucleotide arrays. Specifically, our alms are as follow. 1) To improve on methods for the analysis of raw image data including dynamic gridding and de-blurring, computation of cell intensity and quality scores, background and gradient correction. 2) To design normalization procedures for array-to-array Comparison. 3) To learn and accumulate statistical knowledge about probe-specific effects through the analysis of a large database of expression arrays. This knowledge will be exploited to design methods of analysis that will greatly reduce variability due to probe effects in the estimation and comparison of gene expression levels. This knowledge will also be used to design model- based methods to assess gene presence and differential expression. In addition, probe sequence information will be incorporated into the analysis when it becomes available 4) To develop computer programs based on the above methodological research. Such programs will be made available to the research community to facilitate effective analysis of oligonucleotide expression array data. With current methods of analysis the very high intrinsic sensitivity of oligonucleotide array technology has not been fully realized for gene expression studies. The methods developed in this project will allow improved analysis and will greatly enhance the utility of expression array data.